摘要
传统的网页排序算法只考虑到用户对网页的访问量、网页更新率、网页转载次数等,而忽视了跳出率,跳出率是反映页面流量质量的重要指标.本文将网页跳出因素以权重因子形式融入网页排序Hyperlink-Induced Topic Search(HITS)算法中得到Bounce Rate HITS(BRHITS)算法,更新融入页面跳出率的权威页面鉴别算法的Authority值.实验过程中,利用爬取门户网站数据将HITS算法与基于内容相关性改进的HITS算法(GHITS)、BRHITS算法进行实验对比.实验结果表明,BRHITS算法比上述两种算法的查准率提高10%~30%.因此本文提出的算法能够在一定程度上改善页面排序质量.
For the traditional web page sorting algorithm, it only takes into account the users′ page view, page update rate, and page reprint times, etc.,but ignores the bounce rate, which is an important indicator reflecting the quality of web traffic.In this paper, we consider the page bounce rate factor, and integrate this factor into the Hyperlink-Induced Topic Search(HITS)algorithm in the form of weight factor to get a new Bounce Rate HITS(BRHITS)algorithm, which updates the Authority values of the authoritative page identification algorithm incorporating the page bounce rate.As for the experiments, HITS algorithm is compared with GHITS(improved HITS algorithm based on content relevance) algorithm and BRHITS algorithm by using crawling portal data.The experimental results show that the accuracy of BRHITS algorithm is 10%~30% higher than the above two algorithms.The algorithm proposed in this paper can improve the quality of page sorting to a certain extent.
作者
王嵘冰
刘鹤
WANG Rong-bing;LIU He(School of Information,Liaoning University,Shenyang 110036,China)
出处
《辽宁大学学报（自然科学版）》
CAS
2022年第4期307-313,共7页
Journal of Liaoning University：Natural Sciences Edition
基金
辽宁省社会科学规划基金(L21BGL026)。
